Enriched volatiles and refractories but deficient titanium on the dayside atmosphere of WASP-121b revealed by JWST/NIRISS

TL;DR

This study uses two JWST/NIRISS SOSS secondary eclipses of WASP-121b to infer its dayside atmospheric composition and thermal structure. By employing free, hybrid, and chemical-equilibrium retrievals and explicitly accounting for H2O dissociation and potential reflected light, the authors find a strong dayside thermal inversion ($T$ rising above $3000$ K) with emission features from H2O, CO, VO, and H−, while TiO is strongly depleted, likely due to a nightside cold-trap. The volatile-to-refractory budget is broadly consistent with the stellar value, and the C/O ratio is sensitive to the inclusion of reflected light, ranging from near stellar to super-stellar. The results highlight the need for non-uniform metallicity treatments in equilibrium retrievals and demonstrate how dissociation and Ti-cold trapping shape the observable spectrum, with important implications for planet formation and formation-location in protoplanetary disks.

Abstract

With dayside temperatures elevated enough for all atmospheric constituents to be present in gas form, ultra-hot Jupiters offer a unique opportunity to probe the composition of giant planets. We aim to infer the composition and thermal structure of the dayside atmosphere of the ultra-hot Jupiter WASP-121b from two NIRISS$/$SOSS secondary eclipses observed as part of a full phase curve. We extract the eclipse spectrum of WASP-121b with two independent data reduction pipelines and analyse it using different atmospheric retrieval prescriptions to explore the effects of thermal dissociation, reflected light, and titanium condensation on the inferred atmospheric properties. We find that the observed dayside spectrum of WASP-121b is best fit by atmosphere models possessing a stratospheric inversion with temperatures reaching over 3000K, with spectral contributions from H2O, CO, VO, H-, and either TiO or reflected light. We measure the atmosphere of WASP-121b to be metal enriched (~10x stellar) but comparatively titanium poor (~1x stellar), potentially due to partial cold-trapping. The inferred C/O depends on model assumptions such as whether reflected light is included, ranging from being consistent with stellar if a geometric albedo of zero is assumed to being super-stellar for a freely fitted Ag = 0.16 +/- 0.02. The volatile-to-refractory ratio is measured to be consistent with the stellar value. We infer that WASP-121b has an atmosphere enriched in both volatile and refractory metals, but not in ultra-refractory titanium, suggesting the presence of a nightside cold-trap. Considering H2O dissociation is critical in free retrieval analyses, leading to order-of-magnitude differences in retrieved abundances for WASP-121b if neglected. Simple chemical equilibrium retrievals assuming that all species are governed by a single metallicity parameter drastically overpredict the TiO abundance.

Figures (12)

• Figure 1: Reduction steps and spectral extraction of the NIRISS/SOSS data. Top panel: NIRISS detector after subtraction of the super-bias and non-linearity correction, and ramp-fitting, converted from counts to count rates. Second panel: After bad pixel and cosmic ray hits correction, and subtraction of the non-uniform background. Third panel: Frame after correction of the $1/f$ noise. Bottom panel: Extracted spectra for the first (red) and second (blue) spectral orders.
• Figure 2: Observed NIRISS/SOSS raw white-light phase curve of WASP-121b and model fit. The data are indicated as grey points (50-integration bins shown as white dots for clarity), with the combined astrophysical and systematics model shown in red. The tilt event occurring during the second eclipse (zoom in shown in the inset panel) is marked by the dashed blue line.
• Figure 3: Broadband and spectroscopic secondary eclipses observed as part of the NIRISS/SOSS phase curve of WASP-121b, extracted with the NAMELESS pipeline. The light curve fit (black line) of the broadband order 1 light-curve ($\lambda=0.85-2.83\,\mu$m) is shown, along with the data binned at a resolution of 25 integrations per bin for visual clarity. The 122 light curves of the second order and 361 light curves of the first order are binned into 3 and 6 light curves, respectively (coloured points).
• Figure 4: Extracted dayside planet-to-star flux ratio spectrum of WASP-121b observed with NIRISS SOSS, shown at both native resolution (grey points) and binned at a fixed resolving power of $R$ = 300 (black points). The NIRISS data are in relatively good agreement with previous observations obtained with TESS bourrier_optical_2020, TRAPPIST delrez_wasp-121_2016, HST WFC3/G102 mikal-evans_emission_2019, HST WFC3/G141 mansfield_unique_2021, and 2MASS kovacs_secondary_2019, while also spanning a wider wavelength coverage.
• Figure 5: Results overview for the WASP-121b NIRISS SOSS eclipse hybrid free retrieval including parameterized thermal dissociation. Top left: Retrieved dayside vertical temperature structure (black line: median, grey contours: 1 and 2 $\sigma$ bounds) compared to the average TP profile at mid-eclipse of a fiducial GCM model parmentier_thermal_2018. Top right: Measured dayside brightness temperature of WASP-121b (black points) compared to the best fit atmospheric model (solid green line). Opacity contributions from individual species are shown in colour. Multiple H$_2$O emission bands can be seen, as well as contribution from CO around 2.4 $\mu$m. At shorter wavelengths, H$^{-}$, VO, and TiO are necessary to match the rise in brightness temperature (note that an albedo of zero was assumed for this retrieval). Bottom: Each panel shows the retrieved abundance profiles for an individual species fitted in the retrieval (solid black line: median, grey contours: 1 and 2 $\sigma$ bounds). Average contribution functions (dotted black line) show the pressure level probed by each species. The abundance profiles predicted from an equilibrium chemistry model assuming the median TP profile and a stellar-like composition evans-soma_sio_2025 are also shown for comparison (dashed lime green lines). Constrained abundances are obtained for H$_2$O, CO, VO, TiO, and H$^{-}$, with only upper limits being obtained for the other species. Error bars (1$\sigma$) or upper limits (2$\sigma$) are shown at the approximate average pressure probed (peak of the contribution function).